Machine Learning for the Prediction of Procedural Case Durations Developed Using a Large Multicenter Database: Algorithm Development and Validation Study.

BACKGROUND: Accurate projections of procedural case durations are complex but critical to the planning of perioperative staffing, operating room resources, and patient communication. Nonlinear prediction models using machine learning methods may provide opportunities for hospitals to improve upon current estimates of procedure duration. OBJECTIVE: The aim of this study was to determine whether a machine learning algorithm scalable across multiple centers could make estimations of case duration within a tolerance limit because there are substantial resources required for operating room functioning that relate to case duration. METHODS: Deep learning, gradient boosting, and ensemble machine learning models were generated using perioperative data available at 3 distinct time points: the time of scheduling, the time of patient arrival to the operating or procedure room (primary model), and the time of surgical incision or procedure start. The primary outcome was procedure duration, defined by the time between the arrival and the departure of the patient from the procedure room. Model performance was assessed by mean absolute error (MAE), the proportion of predictions falling within 20% of the actual duration, and other standard metrics. Performance was compared with a baseline method of historical means within a linear regression model. Model features driving predictions were assessed using Shapley additive explanations values and permutation feature importance. RESULTS: A total of 1,177,893 procedures from 13 academic and private hospitals between 2016 and 2019 were used. Across all procedures, the median procedure duration was 94 (IQR 50-167) minutes. In estimating the procedure duration, the gradient boosting machine was the best-performing model, demonstrating an MAE of 34 (SD 47) minutes, with 46% of the predictions falling within 20% of the actual duration in the test data set. This represented a statistically and clinically significant improvement in predictions compared with a baseline linear regression model (MAE 43 min; P<.001; 39% of the predictions falling within 20% of the actual duration). The most important features in model training were historical procedure duration by surgeon, the word "free" within the procedure text, and the time of day. CONCLUSIONS: Nonlinear models using machine learning techniques may be used to generate high-performing, automatable, explainable, and scalable prediction models for procedure duration.

Comparison of serratus anterior plane block with epidural and paravertebral block in critically ill trauma patients with multiple rib fractures.

BACKGROUND: Pain from rib fractures is associated with significant pulmonary morbidity. Epidural and paravertebral blocks (EPVBs) have been recommended as part of a multimodal approach to rib fracture pain, but their utility is often challenging in the trauma intensive care unit (ICU). The serratus anterior plane block (SAPB) has potential as an alternative approach for chest wall analgesia. METHODS: This retrospective study compared critically injured adults sustaining multiple rib fractures who had SAPB (n=14) to EPVB (n=25). Patients were matched by age, body mass index, American Society of Anesthesiology Physical Status, whether the patient required intubation, number of rib fractures and injury severity score. Outcome measures included hospital length of stay, ICU length of stay, preblock and post block rapid shallow breathing index (RSBI) in intubated patients, pain scores and morphine equivalent doses administered 24-hour preblock and post-block in non-intubated patients, and mortality. RESULTS: There were no demographic differences between the two groups after matching. Nearly all of the patients who received either SAPB or EPVB demonstrated a reduction in RSBI or pain scores. The preblock RSBI was higher in the serratus anterior plane block group, but there was no difference between any of the other outcome measures. DISCUSSION: This retrospective study of our institutional data suggests no difference in efficacy between the serratus anterior plane block and neuraxial block for traumatic rib fracture pain in critically ill patients, but the sample size was too small to show statistical equivalence. Serratus anterior plane block is technically easier to perform with fewer theoretical contraindications compared with traditional neuraxial block. Further study with prospective comparative trials is warranted. LEVEL OF EVIDENCE: Retrospective matched cohort; Level IV.

Forecasting a Crisis: Machine-Learning Models Predict Occurrence of Intraoperative Bradycardia Associated With Hypotension.

BACKGROUND: Predictive analytics systems may improve perioperative care by enhancing preparation for, recognition of, and response to high-risk clinical events. Bradycardia is a fairly common and unpredictable clinical event with many causes; it may be benign or become associated with hypotension requiring aggressive treatment. Our aim was to build models to predict the occurrence of clinically significant intraoperative bradycardia at 3 time points during an operative course by utilizing available preoperative electronic medical record and intraoperative anesthesia information management system data. METHODS: The analyzed data include 62,182 scheduled noncardiac procedures performed at the University of Washington Medical Center between 2012 and 2017. The clinical event was defined as severe bradycardia (heart rate <50 beats per minute) followed by hypotension (mean arterial pressure <55 mm Hg) within a 10-minute window. We developed models to predict the presence of at least 1 event following 3 time points: induction of anesthesia (TP1), start of the procedure (TP2), and 30 minutes after the start of the procedure (TP3). Predictor variables were based on data available before each time point and included preoperative patient and procedure data (TP1), followed by intraoperative minute-to-minute patient monitor, ventilator, intravenous fluid, infusion, and bolus medication data (TP2 and TP3). Machine-learning and logistic regression models were developed, and their predictive abilities were evaluated using the area under the ROC curve (AUC). The contribution of the input variables to the models were evaluated. RESULTS: The number of events was 3498 (5.6%) after TP1, 2404 (3.9%) after TP2, and 1066 (1.7%) after TP3. Heart rate was the strongest predictor for events after TP1. Occurrence of a previous event, mean heart rate, and mean pulse rates before TP2 were the strongest predictor for events after TP2. Occurrence of a previous event, mean heart rate, mean pulse rates before TP2 (and their interaction), and 15-minute slopes in heart rate and blood pressure before TP2 were the strongest predictors for events after TP3. The best performing machine-learning models including all cases produced an AUC of 0.81 (TP1), 0.87 (TP2), and 0.89 (TP3) with positive predictive values of 0.30, 0.29, and 0.15 at 95% specificity, respectively. CONCLUSIONS: We developed models to predict unstable bradycardia leveraging preoperative and real-time intraoperative data. Our study demonstrates how predictive models may be utilized to predict clinical events across multiple time intervals, with a future goal of developing real-time, intraoperative, decision support.

Improving Operating Room Efficiency: Machine Learning Approach to Predict Case-Time Duration.

BACKGROUND: Accurate estimation of operative case-time duration is critical for optimizing operating room use. Current estimates are inaccurate and earlier models include data not available at the time of scheduling. Our objective was to develop statistical models in a large retrospective data set to improve estimation of case-time duration relative to current standards. STUDY DESIGN: We developed models to predict case-time duration using linear regression and supervised machine learning. For each of these models, we generated an all-inclusive model, service-specific models, and surgeon-specific models. In the latter 2 approaches, individual models were created for each surgical service and surgeon, respectively. Our data set included 46,986 scheduled operations performed at a large academic medical center from January 2014 to December 2017, with 80% used for training and 20% for model testing/validation. Predictions derived from each model were compared with our institutional standard of using average historic procedure times and surgeon estimates. Models were evaluated based on accuracy, overage (case duration > predicted + 10%), underage (case duration < predicted - 10%), and the predictive capability of being within a 10% tolerance threshold. RESULTS: The machine learning algorithm resulted in the highest predictive capability. The surgeon-specific model was superior to the service-specific model, with higher accuracy, lower percentage of overage and underage, and higher percentage of cases within the 10% threshold. The ability to predict cases within 10% improved from 32% using our institutional standard to 39% with the machine learning surgeon-specific model. CONCLUSIONS: Our study is a notable advancement toward statistical modeling of case-time duration across all surgical departments in a large tertiary medical center. Machine learning approaches can improve case duration estimations, enabling improved operating room scheduling, efficiency, and reduced costs.

Does Attending Surgeon Presence at the Preinduction Briefing Improve Operating Room Efficiency?

OBJECTIVE: To examine if attending surgeon presence at the preinduction briefing is associated with a shorter time to incision. STUDY DESIGN: Retrospective cohort study and survey. SETTING: Tertiary academic medical center. SUBJECTS AND METHODS: A retrospective cohort study was conducted of 22,857 operations by 141 attending surgeons across 12 specialties between August 3, 2016, and June 21, 2018. The independent variable was attending surgeon presence at the preinduction briefing. Linear regression models compared time from room entry to incision overall, by service line, and by surgeon. We hypothesized a shorter time to incision when the attending surgeon was present and a larger effect for cases with complex surgical equipment or positioning. A survey was administered to evaluate attending surgeons' perceptions of the briefing, with a response rate of 68% (64 of 94 attending surgeons). RESULTS: Cases for which the attending surgeon was present at the preinduction briefing had a statistically significant yet operationally minor reduction in mean time to incision when compared with cases when the attending surgeon was absent. After covariate adjustment, the mean time to incision was associated with an efficiency gain of 1.8 ± 0.5 minutes (mean ± SD; P < .001). There were no statistically significant differences in the subgroups of complex surgical equipment and complex positioning or in secondary analysis comparing service lines. The surgeon was the strongest confounding variable. Survey results demonstrated mild support: 55% of attending surgeons highly prioritized attending the preinduction briefing. CONCLUSION: Attending surgeon presence at the preinduction briefing has only a minor effect on efficiency as measured by time to incision.

A randomized comparison of the Endeavor zotarolimus-eluting stent versus the TAXUS paclitaxel-eluting stent in de novo native coronary lesions 12-month outcomes from the ENDEAVOR IV trial.

OBJECTIVES: The ENDEAVOR IV (Randomized Comparison of Zotarolimus-Eluting and Paclitaxel-Eluting Stents in Patients with Coronary Artery Disease) trial evaluated the safety and efficacy of the zotarolimus-eluting stent (ZES) compared with the paclitaxel-eluting stent (PES). BACKGROUND: First-generation drug-eluting stents have reduced angiographic and clinical restenosis, but long-term safety remains controversial. A second-generation drug-eluting stent, which delivers zotarolimus, a potent antiproliferative agent, via a biocompatible phosphorylcholine polymer on a cobalt alloy thin-strut stent has shown promising experimental and early clinical results. METHODS: This is a prospective, randomized (1:1), single-blind, controlled trial comparing outcomes of patients with single de novo coronary lesions treated with ZES or PES. The primary end point was noninferiority of 9-month target vessel failure defined as cardiac death, myocardial infarction, or target vessel revascularization. RESULTS: Among a total of 1,548 patients assigned to ZES (n = 773) or PES (n = 775), at 9 months, ZES was noninferior to PES with rates of target vessel failure 6.6% versus 7.1%, respectively (p(noninferiority) < or = 0.001). There were fewer periprocedural myocardial infarctions with ZES (0.5% vs. 2.2%; p = 0.007), whereas at 12 months, there were no significant differences between groups in rates of cardiac death, myocardial infarction, target vessel revascularization, or stent thrombosis. Although incidence of 8-month binary angiographic in-segment restenosis was higher in patients treated with ZES versus PES (15.3% vs. 10.4%; p = 0.284), rates of 12-month target lesion revascularization were similar (4.5% vs. 3.2%; p = 0.228), especially in patients without planned angiographic follow-up (3.6% vs. 3.2%; p = 0.756). CONCLUSIONS: These findings demonstrate that ZES has similar clinical safety and efficacy compared with PES in simple and medium complexity single de novo coronary lesions. (ENDEAVOR IV Clinical Trial; NCT00217269).

Clinical and angiographic outcomes in diabetics from the ENDEAVOR IV trial: randomized comparison of zotarolimus- and paclitaxel-eluting stents in patients with coronary artery disease.

OBJECTIVES: The aim of this study was to examine outcomes related to the use of the Endeavor zotarolimus-eluting stent (ZES) (Medtronic CardioVascular, Santa Rosa, California) compared with the TAXUS paclitaxel-eluting stent (PES) (Boston Scientific Corp., Natick, Massachusetts) in the 477 patients with diabetes mellitus (DM) enrolled in the randomized ENDEAVOR IV (Randomized Comparison of Zotarolimus- and Paclitaxel-Eluting Stents in Patients with Coronary Artery Disease) trial. BACKGROUND: Percutaneous coronary intervention (PCI) in diabetic patients is associated with increased rates of restenosis-related end points compared with PCI in nondiabetic patients. Although ZES has been associated with similar clinical efficacy compared with PES in the overall trial population of the ENDEAVOR IV trial, whether these results are maintained in the higher-risk restenosis subgroup of patients with DM has not been determined. METHODS: Clinical and angiographic outcomes were compared according to randomized treatment assignment to either ZES or PES. RESULTS: Baseline characteristics were similar among ZES (n = 241) and PES (n = 236) diabetic patients, with slightly longer lesion lengths in PES-treated patients (12.9 mm vs. 14.0 mm, p = 0.041). Among the 86 DM patients assigned to routine angiographic follow-up (18% of the overall DM cohort), in-stent percent diameter stenosis at 8 months was greater among ZES-treated patients (32.9 vs. 21.1, p = 0.023), with a trend toward higher in-stent late loss. One-year clinical outcomes were similar among DM patients treated with either ZES or PES (target vessel failure: 8.6% vs. 10.8%, p = 0.53; target lesion revascularization: 6.9% vs. 5.8%, p = 0.70; target vessel revascularization: 8.6% vs. 9.4%, p = 0.87). There were no significant interactions between DM status and stent type with respect to the outcomes measured, and the relative efficacy/safety of ZES and PES were similar among insulin- and noninsulin-requiring subgroups. CONCLUSIONS: One-year clinical outcomes were similar among DM patients treated with ZES and PES in the ENDEAVOR IV trial. These findings parallel the overall trial results, which demonstrated similar efficacy and safety of ZES and PES for single de novo coronary lesions.

Late safety, efficacy, and cost-effectiveness of a zotarolimus-eluting stent compared with a paclitaxel-eluting stent in patients with de novo coronary lesions: 2-year follow-up from the ENDEAVOR IV trial (Randomized, Controlled Trial of the Medtronic Endeavor Drug [ABT-578] Eluting Coronary Stent System Versus the Taxus Paclitaxel-Eluting Coronary Stent System in De Novo Native Coronary Artery Lesions).

OBJECTIVES: The aim of this study was to assess, after 2 years of follow-up, the safety, efficacy, and cost-effectiveness of a zotarolimus-eluting stent (ZES) compared with a paclitaxel-eluting stent (PES) in patients with native coronary lesions. BACKGROUND: Early drug-eluting stents were associated with a small but significant incidence of very late stent thrombosis (VLST), occurring >1 year after the index procedure. The ZES has shown encouraging results in clinical trials. METHODS: The ENDEAVOR IV trial (Randomized, Controlled Trial of the Medtronic Endeavor Drug [ABT-578] Eluting Coronary Stent System Versus the Taxus Paclitaxel-Eluting Coronary Stent System in De Novo Native Coronary Artery Lesions), a randomized (1:1), single-blind, controlled trial (n = 1,548) compared ZES versus PES in patients with single de novo coronary lesions. Two-year follow-up was obtained in 96.0% of ZES and 95.4% of PES patients. The primary end point was target vessel failure (TVF), and safety end points included Academic Research Consortium-defined stent thrombosis. Economic end points analyzed included quality-adjusted survival, medical costs, and relative cost-effectiveness of ZES and PES. RESULTS: The TVF at 2 years was similar in ZES and PES patients (11.1% vs. 13.1%, p = 0.232). There were fewer myocardial infarctions (MIs) in ZES patients (p = 0.022), due to fewer periprocedural non-Q-wave MIs and fewer late MIs between 1 and 2 years. Late MIs were associated with increased VLST (PES: 6 vs. ZES: 1; p = 0.069). Target lesion revascularization was similar comparing ZES with PES (5.9% vs. 4.6%; p = 0.295), especially in patients without planned angiographic follow-up (5.2% vs. 4.9%; p = 0.896). The cost-effectiveness of ZES and PES was similar. CONCLUSIONS: After 2 years of follow-up, ZES demonstrated efficacy and cost-effectiveness comparable to PES, with fewer MIs and a trend toward less VLST. (The ENDEAVOR IV Clinical Trial: A Trial of a Coronary Stent System in Coronary Artery Lesions; NCT00217269).

Initial impact of drug-eluting stent use on the spectrum of patients undergoing percutaneous coronary intervention.

BACKGROUND: The availability of drug-eluting stents was expected to expand the application of percutaneous coronary intervention to a wider group of patients and lesion types. We sought to determine whether drug-eluting stents' availability has changed the practices of operators performing percutaneous coronary intervention with regard to patient selection and procedural factors. METHODS: We compared the clinical and angiographic characteristics of patients who underwent percutaneous coronary intervention at the Methodist Hospital, Houston, Texas, during three periods: June 2002-March 2003, June 2003-March 2004 and June 2004-March 2005. Sirolimus-eluting stents were available during the latter two periods. Paclitaxel eluting stents became available during the third period. RESULTS: A greater proportion of patients undergoing diagnostic catheterization were referred for percutaneous coronary intervention during the latter two periods (26.8%, 30.4%, 30.4%, respectively, P<0.0001). Patients undergoing stent placement during the latter two periods were more likely to have hypertension, hyperlipidemia or to have undergone previous percutaneous coronary intervention. The average implanted stent length was greater during the latter periods (14.5+/-4.6, 16.4+/-5.2, 16.8+/-5.5 mm, respectively, P<0.0001) and the average stent diameter was smaller (3.23+/-1.22, 3.13+/-1.16, 3.02+/-0.6 mm, respectively, P<0.0001). The frequency of percutaneous coronary intervention involving long lesions (>20 mm) and Left Anterior Descending (LAD) lesions was higher in the latter two periods and the frequency of multivessel stenting was higher in the last period. CONCLUSIONS: The spectrum of patients and coronary lesions that have undergone stenting has changed, particularly in the third period when both drug-eluting stent types were available. We observed a gradual shift toward higher-risk clinical and lesion characteristics following the introduction of drug-eluting stents.

Initial experience with a magnetic navigation system for percutaneous coronary intervention in complex coronary artery lesions.

OBJECTIVES: The aim of this study was to evaluate the feasibility of a magnetic-assisted navigation system during percutaneous coronary intervention (PCI) of tortuous and severely angulated coronary arteries. BACKGROUND: The magnetic navigation system consists of two 0.8-T permanent magnets which generate a magnetic field over the heart. Altering the magnetic vector deflects a coronary guidewire with a magnetic tip. METHODS: Patients were selected for magnetic-assisted intervention (MAI) for potentially difficult to cross lesions. The time required for placement of the guidewire, total procedure time, fluoroscopy time, and amount of contrast for the procedure were recorded. There were a total of 59 patients undergoing PCI of 68 lesions. RESULTS: Patients were grouped based on whether MAI was attempted as a first option ("primary attempt"; n = 46) or following failure to pass a conventional guidewire ("secondary attempt"; n = 13). The target lesion was successfully crossed in 49 of 55 lesions (89%) and 9 of 13 lesions (69%) in patients undergoing primary and secondary attempts, respectively. The procedural success rates were 84% and 62%, respectively. Most lesions were located in the circumflex artery territory (39% and 62% of lesions, respectively). The median (25th and 75th percentiles) time for crossing the lesion was longer in the secondary attempt group (14.8 [5, 15.5] vs. 28.9 [8, 38] min). Median fluoroscopy time and median contrast used were also higher among the secondary attempt group. CONCLUSIONS: This first report of MAI suggests that it may become a useful adjunct for wire placement in difficult coronary interventions.

Pharmacodynamic profile of the direct thrombin antagonist bivalirudin given in combination with the glycoprotein IIb/IIIa antagonist eptifibatide.

BACKGROUND: Because of the adverse characteristics associated with heparin, direct antagonists of thrombin have been investigated as anticoagulants during percutaneous coronary interventions. However, the hematologic and clinical interactions between direct thrombin antagonists and inhibitors of platelet glycoprotein IIb-IIIa are incompletely explored. METHODS: Forty-two patients who underwent elective percutaneous coronary intervention were randomized to receive a bivalirudin 1.0 mg/kg bolus followed by a 4-hour infusion at 2.5 mg/kg/h; a bivalirudin 0.75 mg/kg bolus followed by a 4-hour infusion at 1.75 mg/kg; or a heparin 60 U/kg bolus. All the patients also received eptifibatide, given as 2 sequential boluses of 180 microg/kg followed by a 2 microg/kg/min infusion for 18 to 24 hours, and aspirin. RESULTS: After the bolus dose of the study drug, turbidimetric platelet aggregation in response to 5 micromol/L adenosine diphosphate increased in patients assigned to heparin but not those assigned to bivalirudin. After eptifibatide, platelet aggregation was eliminated in all 3 treatment groups. The effect of heparin and the effects of both bivalirudin regimens on the formation of thrombin antithrombin complexes and prothrombin fragment 1.2 were comparable. Neither agent affected the formation of platelet-monocyte complexes or expression of CD 63 lysosomal antigen. There were no major bleeding events, and a single non-Q-wave myocardial infarction (MI) occurred in a patient treated with bivalirudin. CONCLUSION: These findings show the feasibility of combining the direct thrombin antagonist bivalirudin with a potent antagonist of platelet glycoprotein IIb-IIIa. Clinical trials are needed to assess the safety and efficacy of this combination.

CARDIOGENIC ACUTE RENAL FAILURE (CARF) FOLLOWING OPEN-HEART SURGERY.

Although previous reports have attributed acute renal failure (ARF) following cardiovascular surgery to acute tubular necrosis (ATN), little emphasis has been placed on renal failure due to congestive heart failure (CARF). Of 100 cases of ARF studied prospectively over an 18-month period, 36 occurred after open-heart surgery. Nineteen of these cases were associated with heart failure. The remaining 17 had ATN as manifested by high urinary sodium, low urine/plasma creatinine, and abnormal urinary sediment. At the onset of CARF, intravascular volume expansion was universally present, and oliguria with pulmonary edema was common. Urinary chemistries were (mean +/- SD): sodium (mEq/L) 8 +/- 7, U/P creatinine 72 +/- 45, and FENa (%) 0.1 +/- 0.1. Therapy consisted of digoxin, furosemide (F), vasopressors (V), and, when indicated, intraaortic balloon counterpulsation. Survivors of CARF responded more frequently to F and required less V. Ultimately, survival depended upon improvement in cardiac performance. All oliguric ATN patients failed to respond to F. Mortality for the CARF group was 52%. In contrast, 82% of the oliguric ATN group expired, whereas overall ATN mortality was 60%. Cardiogenic acute renal failure is a frequent cause of ARF after open-heart surgery in our institution. It is characterized by prerenal urinary chemistries, has a high mortality, and may be reversible.